CN103488485A - Parallel migration method for optimizing operating performance of application software on server - Google Patents

Parallel migration method for optimizing operating performance of application software on server Download PDF

Info

Publication number
CN103488485A
CN103488485A CN201310438918.XA CN201310438918A CN103488485A CN 103488485 A CN103488485 A CN 103488485A CN 201310438918 A CN201310438918 A CN 201310438918A CN 103488485 A CN103488485 A CN 103488485A
Authority
CN
China
Prior art keywords
parallel
implantation method
migration method
parallel migration
application software
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN201310438918.XA
Other languages
Chinese (zh)
Inventor
胡自玉
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Inspur Electronic Information Industry Co Ltd
Original Assignee
Inspur Electronic Information Industry Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Inspur Electronic Information Industry Co Ltd filed Critical Inspur Electronic Information Industry Co Ltd
Priority to CN201310438918.XA priority Critical patent/CN103488485A/en
Publication of CN103488485A publication Critical patent/CN103488485A/en
Pending legal-status Critical Current

Links

Landscapes

  • Stored Programmes (AREA)

Abstract

The invention provides a parallel migration method for optimizing the operating performance of application software on a server. The parallel migration method re-establishes and migrates density functional energy terms and optimizes and analyzes a density functional theory in a parallel migration mode so as to achieve the purposes of software optimization and quantification. By adopting the parallel migration method, the space of the application field can be effectively widened, the limitation of a single optimizer is avoided, and accordingly resources of clusters and a super computer are effectively utilized. The parallel migration method has considerable effectiveness on the simple optimization of library functions and instruction sets. By means of the parallel migration method, the availability of the density functional theory on computational chemistry, material science and nanoscience is accelerated, the density functional theory can be conveniently and effectively optimized and analyzed, and the study efficiency in the fields of the computational chemistry, the material science, the nanoscience and the like can be improved.

Description

A kind of parallel implantation method of optimizing application software runnability on server
Technical field
This present invention relates to high-performance computing sector in chemistry, the application of material science and field of nanometer technology, concrete a kind of method of parallel transplanting that adopts is carried out establishment again and the transplanting of Density functional energy term, density functional theory is optimized and analyzes and then reaches and optimize the purpose that quantizes software.
Background technology
Along with the continuous progress of science, people are to the cognition of material and the level that extends to electronics.The foundation of schrodinger equation makes intramundane Theoretical Framework set up fully.By quantum-mechanical means, people can accurately hold the form of object on electronics, atom or molecular scale.But middle difficulty is the problem that how to solve the schrodinger equation that is deep into electronic structure.So far, about the theory of computation of electronic structure, mainly contain two kinds, a kind of is to be based upon other theoretical model that the quantification theory from the beginning calculated on Hartree good fortune gram (HF) theoretical foundation or the HF of take are benchmark.Can obtain other character of energy and the material of arbitrary accuracy by the calculating of these theories.In addition, Hartree good fortune gram (HF) and other theory relevant with it can system the degree of accuracy to calculating corrected, but the calculated amount of this theoretical method is large especially, can consume huge computational resource, outer-shell electron also can only be considered 4 heavy rail roads of some atoms.The computing method of another kind of electronic structure are density functional theory (DFT).Its major advantage with respect to the HF theory is that can to expand population to be N for the system calculated 3system.In the DFT theory, energy can separate.Its energy comprises non-interacting kinetic energy, the interactional energy of electronics and atomic nucleus, the potential energy of electronics and electron interaction, and exchange correlation energy.Theoretical main employing of current DFT " Jacob ladder technology " is devoted to better more detailed description exchange correlation functional item of research.
First three in a density functional theory energy can be used for analyzing and upgrading the exchange correlation item, and the exchange correlation item, and the calculating of the differential of associations must rely on a series of grid.Even the calculating of the exchange correlation formed by density afterwards energy also just can be carried out on the system network lattice point, a series of grid does not possess the analytical calculation of arbitrary accuracy certainly, but calculate, depends on complicated grid.And a series of energy in computation process on grid calculates and depends on the direction of molecule with respect to grid, this does not obviously have any physical significance.Add, carry out again improving and can realize of grid, and then can shorten the speed of calculating.
For the arithmetic speed of software on server that improves the DFT theory, designed a kind of parallel implantation method here and optimized and analyzed the scheme of DFT software, and obtained desirable result.This parallel implantation method has many advantages, and its main advantage is scalability, and this has also broken a simple optimizer and may be trapped in minimum range of application and can not find better available point.The parallel implantation method here is the field that simple implantation method has been widened application, can more conveniently find the parameter of the DFT software that can optimize , accelerate the speed that software is restrained.
Summary of the invention
The purpose of this invention is to provide a kind of parallel implantation method of optimizing application software runnability on server.
The objective of the invention is to realize in the following manner, the Optimal Parameters formula is as follows,
=
Figure 201310438918X100002DEST_PATH_IMAGE003
Figure 201310438918X100002DEST_PATH_IMAGE005
Figure 826036DEST_PATH_IMAGE006
Proofread and correct optimum absolute average error
Figure 201310438918X100002DEST_PATH_IMAGE007
, obtain other relative average error
Figure 153113DEST_PATH_IMAGE008
, these data determine to back realize that the capable of parallel moving method for planting is marginal, can realize the Fast Convergent of single implantation method, and on the result calculated without impact, basic program architecture and organization flow are:
1) test one by one the Optimal Parameters of every kind of element in the theoretical software of the corresponding DFT of single implantation method
Figure 265161DEST_PATH_IMAGE001
;
2) 0.5
Figure 201310438918X100002DEST_PATH_IMAGE009
, test and find the optimized parameter of corresponding every kind of element value;
3) calculate atomic
Figure 440108DEST_PATH_IMAGE010
the derivative of absolute average error
Figure 824953DEST_PATH_IMAGE002
;
4) proofread and correct optimum absolute average error
Figure 827675DEST_PATH_IMAGE007
, can obtain other relative average error
Figure 727498DEST_PATH_IMAGE008
, these numerical value are the initial optimization parameter in parallel implantation method;
5) the implantation method program Python of parallel version is installed;
6) realize the combination of DFT application software and implantation method program Python;
7) find the optimization parameters of the optimum of each element
Figure 201310438918X100002DEST_PATH_IMAGE011
.
the invention has the beneficial effects as follows: the present invention is directed to single method application of transplanting is limited and the runnability of application software on server is not high and carry out some new technological improvements and invention.Present technique invention mainly can be good at optimizing on the basis of software of application of DFT theory at single implantation method, adopted parallel method singly to transplant optimizer and become parallel optimizer, greatly improved its range of application and also reached the performance that effective optimization application software is moved on server.
Embodiment
The invention describes a kind of parallel method acceleration of transplanting and availability of Optimal Density Functional Theory utilized, it is characterized in that: the space in effective broaden application field, thereby the more efficiently resource of utilizing cluster and supercomputer, the method can effectively reduce the data traffic when quantizing software and moving on server in calculating process, thus more efficiently a plurality of cores of utilizing on a plurality of clustered nodes.
In order to make purpose of the present invention, technical scheme and advantage more clear, we use the guassian application software for example, the committed step in the present invention are elaborated, for realizing that the parameter that optimization is set provides in detail.
The Optimal Parameters formula is as follows,
Figure 168974DEST_PATH_IMAGE002
=
Figure 470643DEST_PATH_IMAGE003
Proofread and correct optimum absolute average error
Figure 460574DEST_PATH_IMAGE007
, can obtain other relative average error
Figure 413486DEST_PATH_IMAGE008
.These data determine to back realize that the capable of parallel moving method for planting is marginal, can realize the Fast Convergent of single implantation method, and on the result calculated without impact.
Basic program architecture and organization flow are:
1. test one by one the Optimal Parameters of every kind of element in the theoretical software of the corresponding DFT of single implantation method
Figure 23590DEST_PATH_IMAGE001
;
2. 0.5
Figure 694743DEST_PATH_IMAGE009
, test and find the optimized parameter of corresponding every kind of element
Figure 868236DEST_PATH_IMAGE001
value;
3. calculate atomic
Figure 691966DEST_PATH_IMAGE010
the derivative of absolute average error
Figure 722239DEST_PATH_IMAGE002
;
4. proofread and correct optimum absolute average error
Figure 818371DEST_PATH_IMAGE007
, can obtain other relative average error , these numerical value are the initial optimization parameter in parallel implantation method;
5. the implantation method program Python of parallel version is installed;
6. realize the combination of DFT application software and implantation method program Python;
7. find the optimization parameters of the optimum of each element
Figure 270529DEST_PATH_IMAGE011
Parallel implantation method of the present invention can accelerate the optimization to the DFT application software greatly, for using the guassian application software, it is example, found its Optimal Parameters, relatively be easy to realize that its optimization arranges, can utilize greatly the scientific research of limited computational resource speed-up computation chemistry, material science and nano science etc., and the loss that can reduce server improves the server operation performance, goes back relative energy-saving.Application example Gaussian application software Optimal Parameters numerical value of the present invention is as following table:
Element PGA
H 0.758
Li 0.565
Be 0.879
B 0.617
C 0.667
N 0.665
O 0.645
F 0.613
Na 0.545
Al 0.754
Si 0.523
P 0.813
S 0.705
The present invention has fully dissected the theoretical operation characteristic of DFT application software, has utilized fully the program Python that can realize parallel implantation method, has significantly accelerated the travelling speed on server of application software.The method is that a kind of application software that solves is moved effective means consuming time at server, and the popularization that relatively is widely used, greatly facilitate the researchist micro-, the system of receiving are done to more deep exploration and discovery.
Except the described technical characterictic of instructions, be the known technology of those skilled in the art.

Claims (1)

1. a parallel implantation method of optimizing application software runnability on server, is characterized in that the Optimal Parameters formula is as follows,
Figure 201310438918X100001DEST_PATH_IMAGE001
=
Figure 957878DEST_PATH_IMAGE002
Figure 983602DEST_PATH_IMAGE004
Figure 201310438918X100001DEST_PATH_IMAGE005
Proofread and correct optimum absolute average error , obtain other relative average error
Figure 201310438918X100001DEST_PATH_IMAGE007
, these data determine to back realize that the capable of parallel moving method for planting is marginal, can realize the Fast Convergent of single implantation method, and on the result calculated without impact, basic program architecture and organization flow are:
1) test one by one the Optimal Parameters of every kind of element in the theoretical software of the corresponding DFT of single implantation method ;
2) 0.5
Figure 201310438918X100001DEST_PATH_IMAGE009
, test and find the optimized parameter of corresponding every kind of element
Figure 673407DEST_PATH_IMAGE008
value;
3) calculate atomic
Figure 338875DEST_PATH_IMAGE010
the derivative of absolute average error
Figure 93204DEST_PATH_IMAGE001
;
4) proofread and correct optimum absolute average error
Figure 892533DEST_PATH_IMAGE006
, can obtain other relative average error
Figure 619181DEST_PATH_IMAGE007
, these numerical value are the initial optimization parameter in parallel implantation method;
5) the implantation method program Python of parallel version is installed;
6) realize the combination of DFT application software and implantation method program Python;
7) find the optimization parameters of the optimum of each element
Figure DEST_PATH_IMAGE011
.
CN201310438918.XA 2013-09-25 2013-09-25 Parallel migration method for optimizing operating performance of application software on server Pending CN103488485A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201310438918.XA CN103488485A (en) 2013-09-25 2013-09-25 Parallel migration method for optimizing operating performance of application software on server

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201310438918.XA CN103488485A (en) 2013-09-25 2013-09-25 Parallel migration method for optimizing operating performance of application software on server

Publications (1)

Publication Number Publication Date
CN103488485A true CN103488485A (en) 2014-01-01

Family

ID=49828742

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201310438918.XA Pending CN103488485A (en) 2013-09-25 2013-09-25 Parallel migration method for optimizing operating performance of application software on server

Country Status (1)

Country Link
CN (1) CN103488485A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105260169A (en) * 2014-07-16 2016-01-20 中兴通讯股份有限公司 Cross-platform python program transplanting method and device

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20040107056A1 (en) * 2002-05-28 2004-06-03 Robert Doerksen Methods, systems, and computer program products for computational analysis and design of amphiphilic polymers
CN102937946A (en) * 2012-10-30 2013-02-20 复旦大学 Complicated function minimal value searching method based on constrained regular pattern

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20040107056A1 (en) * 2002-05-28 2004-06-03 Robert Doerksen Methods, systems, and computer program products for computational analysis and design of amphiphilic polymers
CN102937946A (en) * 2012-10-30 2013-02-20 复旦大学 Complicated function minimal value searching method based on constrained regular pattern

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
LILIN LU.ET: "《An improved B3LYP method in the calculation of organic thermochemistry and reactivity》", 《COMPUTATIONAL AND THEORETICAL CHEMISTRY》 *
张颖等: "《新一代密度泛函方法XYG3》", 《化学进展》 *

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105260169A (en) * 2014-07-16 2016-01-20 中兴通讯股份有限公司 Cross-platform python program transplanting method and device

Similar Documents

Publication Publication Date Title
CN109902818A (en) A kind of distributed accelerated method and system towards deep learning training mission
Arita et al. Large-scale DFT simulations with a linear-scaling DFT code CONQUEST on K-computer
CN102692491B (en) Soil moisture characteristic parameter calculating method based on a staging tabu searching algorithm
Surmin et al. Co-design of a particle-in-cell plasma simulation code for Intel Xeon Phi: a first look at Knights Landing
CN108958852A (en) A kind of system optimization method based on FPGA heterogeneous platform
CN102937946B (en) A kind of complicated function minimal value searching method based on constraint regular pattern
CN103198167B (en) The computing method of semiconductor-quantum-point equilibrium composition
CN103488485A (en) Parallel migration method for optimizing operating performance of application software on server
Cai et al. A parallel node-based solution scheme for implicit finite element method using GPU
CN102779025A (en) Parallel PLSA (Probabilistic Latent Semantic Analysis) method based on Hadoop
CN109408838B (en) Method and system for rapidly analyzing residual oil of fracture-cavity oil reservoir
Eisenbach et al. A scalable method for ab initio computation of free energies in nanoscale systems
Bigaouette et al. Nonlinear grid mapping applied to an FDTD-based, multi-center 3D Schrödinger equation solver
Wei et al. GPU acceleration of a 2D compressible Euler solver on CUDA-based block-structured Cartesian meshes
Chen et al. Lattice Boltzmann method for multiscale self-consistent field theory simulations of block copolymers
CN102867125A (en) Method for calculating dynamic time step of molecular dynamics
Mou An efficient ant colony system for solving the new generalized traveling salesman problem
Song et al. Hybrid particle swarm cooperative optimization algorithm and its application to MBC in alumina production
Wang et al. Computer program for directed structure topology optimization
Liu et al. Asynchronous Stochastic Gradient Descent for Extreme-Scale Recommender Systems
Liu et al. Exploiting sample diversity in distributed machine learning systems
Fu et al. Unleashing the performance potential of CPU-GPU platforms for the 3D atmospheric Euler solver
Li et al. Pplsa: Parallel probabilistic latent semantic analysis based on mapreduce
Hongli et al. Accelerated PSO Enhanced by Henon-chaos to Resolve Function Optimization Problems [J]
CN102663141A (en) Multi-channel quantification and hierarchical clustering method based on multi-core parallel computation

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
WD01 Invention patent application deemed withdrawn after publication
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20140101